Unbiased Identification of Functional Barrier Insulators in Primary Human Erythroid Cells,

Abstract 3385

Barrier insulators function to actively maintain the boundaries between heterochromatin and euchromatin. They are critical for regulation of cell-type specific gene expression in normal development and differentiation. Mutations that disrupt barrier insulator function have been associated with developmental disorders, malignancies, and inherited hemolytic anemias. Barrier insulators are poorly understood in mammalian cells, with much of the available data coming from model organisms. In vertebrates, the best characterized barrier insulator is the 5' hypersensitive site in the LCR of the chicken β-globin gene cluster (cHS4). In cHS4, barrier insulator function is mediated by binding of the upstream stimulatory factor (USF) proteins, which bind specific DNA sequences and recruit multiple regulatory proteins, including histone aceytltranserases (HATs) and histone methyltransferases (MTs), which maintain DNA in a euchromatin state. The cHS4 barrier also recruits the protein VEZF1, recently shown to mediate protection of DNA from methylation. We hypothesize that there is a common regulatory signature for cell-type specific barrier insulators characterized by binding of the USF proteins, with recruitment of HATs, MTs, and other proteins in the genome of human erythroid cells. To test this hypothesis, we utilized chromatin immunoprecipitation coupled with massively parallel sequencing (ChIP-seq) to generate genome-wide maps of barrier-associated proteins and histone modifications in primary human erythroid cells (R3/R4 stage). Regions where barrier-associated proteins co-localize, representing potential barrier insulators, were identified then subjected to functional analysis in position effect variegation (PEV) assays. Genome-wide, 3825 sites bound the USF proteins USF1 and USF2 with their associated MTs (PRMT1/PRMT4), and HATs (P300, PCAF, SRC1). The genome-wide binding of VEZF1 was compared to the binding of the USFs, MTs, and HATs. VEZF1 bound 1129 (30%) of the potential barrier sites. The role of CTCF in barrier insulators is controversial. It is dispensable for cHS4 barrier function in chicken erythroid cells, but in human cells, it marks chromatin boundaries in a cell-type specific manner. CTCF ChIP-seq in erythroid cells revealed that a very large number of the barrier-associated sites (3382, 88%) bound CTCF. Together, 1167 sites bound all 9 factors. These sites were located primarily in gene promoters (42%) and 5' untranslated regions (23%), consistent with data from Drosophila, where barriers are frequently associated with gene promoters. Active barriers are associated with an “open” chromatin structure and lack CpG methylation, thus these epigenetic marks were assessed at the predicted barrier sites. The majority of sites, 96%, had the active histone mark H3K4me2, while only 0.02% were positive for the repressive histone mark H3K27me3. To assess CpG methylation, methyl binding domain pull down was coupled with massively parallel sequencing (MethylSeq). 3676 regions of CpG methylation were identified, but none overlapped with the barrier signature. PEV assays, which assesses the ability of a region of DNA to protect a reporter gene from heterochromatin-mediated silencing, were used to determine if selected sites identified by ChIP-seq studies had barrier insulator function in vivo. Constructs containing an EF1alpha promoter directing an EGFP reporter gene-IRES-hygromycin cassette were flanked by potential barriers and stably transfected into K562 cells. Results from single copy clones were normalized to the cHS4 positive control. Sites tested included an intergenic site on chromosome 11 located >100kb from any known gene (site 1), which bound the USFs, PRMTs, PCAF, SRC1, and CTCF, and a site in intron 1 of the band 3 gene (site 2), which bound the USFs, PRMT4, P300, PCAF, and SRC1. Both sites were shown to have barrier activity (site 1 x²= 6.77, p<0.01 and site 2 x²= 3.30, p<0.06), demonstrating that our molecular signature can predict functional barrier insulators. The orientation dependence of vertebrate barrier elements has never been described. When site 1 and 2 were analyzed in the opposite orientation relative to the direction of transcription, neither had barrier function. Unbiased identification of barrier insulators on a genome wide scale will provide novel insights into normal erythropoiesis and its perturbation in human disease.